Fix bug in metric computation when there is overlap with grid objective

[desilinguist]

This PR addresses #564.

• Moves the overlap checking from the configuration level (in config.py) to the experiment level (in learner.py).
• Adds another check for code added in the earlier metric computation PR: the message about probabilities being used to compute the metric is now not shown when computing the objective since doing so might send conflicting messages to the user given the overlap message.
• Remove old test for overlap checking from test_input.py
• Add new test for overlap checking to test_classification.py.

[pep8speaks]

Hello @desilinguist! Thanks for updating this PR.

• In the file tests/test_classification.py:

Line 1739:101: E501 line too long (101 > 100 characters)

Comment last updated at 2019-10-22 00:57:35 UTC

[codecov]

Codecov Report

Merging #567 into master will increase coverage by 3.32%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #567      +/-   ##
==========================================
+ Coverage   91.69%   95.02%   +3.32%     
==========================================
  Files          20       20              
  Lines        2975     2972       -3     
==========================================
+ Hits         2728     2824      +96     
+ Misses        247      148      -99

Impacted Files	Coverage Δ
skll/config.py	95.12% <ø> (+1.18%)	⬆️
skll/learner.py	96% <100%> (+7.28%)	⬆️
skll/metrics.py	97.14% <0%> (+2.85%)	⬆️
skll/experiments.py	95.52% <0%> (+5.8%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e6da3ca...ca539b1. Read the comment docs.